Latent Porosity in Alkali-Metal M2B12F12 Salts ... - ACS Publications

Sep 21, 2017 - Matthew R. Lacroix,. †. Hui Wu,. §. Wei Zhou,. §. W. Matthew Jones,. † ...... (15) Glasser, L.; Jones, F. Systematic Thermodynami...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/IC

Latent Porosity in Alkali-Metal M2B12F12 Salts: Structures and Rapid Room-Temperature Hydration/Dehydration Cycles Dmitry V. Peryshkov,*,†,‡ Eric V. Bukovsky,*,† Matthew R. Lacroix,† Hui Wu,§ Wei Zhou,§ W. Matthew Jones,† Matic Lozinšek,†,⊥ Travis C. Folsom,† D. Luke Heyliger,† Terrence J. Udovic,§ and Steven H. Strauss*,† †

Department of Chemistry, Colorado State University, Fort Collins, Colorado 80523, United States Department of Chemistry and Biochemistry, University of South Carolina, Columbia, South Carolina 29208, United States § Center for Neutron Research, National Institute of Standards and Technology (NIST), Gaithersburg, Maryland 20899, United States ⊥ Department of Inorganic Chemistry and Technology, Jožef Stefan Institute, 1000 Ljubljana, Slovenia ‡

S Supporting Information *

ABSTRACT: Structures of the alkali-metal hydrates Li2(H2O)4Z, LiK(H2O)4Z, Na2(H2O)3Z, and Rb2(H2O)2Z, unit cell parameters for Rb2Z and Rb2(H2O)2Z, and the density functional theory (DFT)-optimized structures of K2Z, K2(H2O)2Z, Rb2Z, Rb2(H2O)2Z, Cs2Z, and Cs2(H2O)Z are reported (Z2− = B12F122−) and compared with previously reported X-ray structures of Na2(H2O)0,4Z, K2(H2O)0,2,4Z, and Cs2(H2O)Z. Unusually rapid roomtemperature hydration/dehydration cycles of several M2Z/M2(H2O)nZ salt hydrate pairs, which were studied by isothermal gravimetry, are also reported. Finely ground samples of K2Z, Rb2Z, and Cs2Z, which are not microporous, exhibited latent porosity by undergoing hydration at 24−25 °C in the presence of 18 Torr of H2O(g) to K2(H2O)2Z, Rb2(H2O)2Z, and Cs2(H2O)Z in 18, 40, and 16 min, respectively. These hydrates were dehydrated at 24−25 °C in dry N2 to the original anhydrous M2Z compounds in 61, 25, and 76 min, respectively (the exact times varied from sample to sample depending on the particle size). The hydrate Na2(H2O)2Z also exhibited latent porosity by undergoing multiple 90 min cycles of hydration to Na2(H2O)3Z and dehydration back to Na2(H2O)2Z at 23 °C. For the K2Z, Rb2Z, and Cs2Z transformations, the maximum rate of hydration (rhmax) decreased, and the absolute value of the maximum rate of dehydration (rdmax) increased, as T increased. For K2Z ↔ K2(H2O)2Z hydration/dehydration cycles with the same sample, the ratio rhmax/rdmax decreased 26 times over 8.6 °C, from 3.7 at 23.4 °C to 0.14 at 32.0 °C. For Rb2Z ↔ Rb2(H2O)2Z cycles, rhmax/rdmax decreased from 0.88 at 23 °C to 0.23 at 27 °C. For Cs2Z ↔ Cs2(H2O)Z cycles, rhmax/rdmax decreased 20 times over 8 °C, from 6.7 at 24 °C to 0.34 at 32 °C. In addition, the reversible substitution of D2O for H2O in fully hydrated Rb2(H2O)2Z in the presence of N2/16 Torr of D2O(g) was complete in only 60 min at 23 °C. frameworks (MOFs)22 and zeolites;23−26 (iv) current interest in the superconductivity in hydrated NaxCoO2,27 superionic conductivity in anhydrous Na2B12H12,28,29 Li+ ion conductivity in hydrated Li0.5FeOCl,30 and the superior thermoelectric response of hydrated NaxRhO2;31 (v) extensive dehydration/ rehydration studies of pharmaceutical and antimicrobial hydrates.32−34 In 2010, we reported that the solid-state transformations K2Z + 2H2O(g) → K2(H2O)2Z and K2(H2O)2Z → K2Z + 2H2O(g) under 21 and 0 Torr of partial pressure of H2O(g) in He, respectively, required only ca. 15 and 75 min, respectively, at a constant temperature (T) of 25 °C (>98% complete reactions; samples were finely ground microcrystalline powders).1

1. INTRODUCTION We are studying anhydrous, hydrated, and solvated salts of the icosahedral superweak anion B12F122− (hereinafter Z2−).1−8 Our interests are multifold: (i) structural changes that occur upon reversible solid-state hydration/dehydration or solvation/ desolvation in the presence/absence of solvent vapor can serve as models for the solid-state diffusion of gaseous reactants and products in lattices;9−13 (ii) thermodynamic and kinetic studies of metal salt hydration/dehydration may help to improve the efficacy of metal salt hydrate pairs used for the storage of low potential heat from solar energy (there are many thermodynamic studies but very few kinetic dehydration/rehydration studies and none at temperatures ≤32 °C);14−17 (iii) the structures of salt hydrates can serve as models for the hydration of metal ions or metal cation−anion inner-sphere ion pairs in solution18−21 or in extended solids such as metal−organic © 2017 American Chemical Society

Received: August 11, 2017 Published: September 21, 2017 12023

DOI: 10.1021/acs.inorgchem.7b02081 Inorg. Chem. 2017, 56, 12023−12041

Article

Inorganic Chemistry

This does not imply the recommendation or endorsement of these suppliers by NIST.] High-Purity Anhydrous Li2Z and Li2(H2O)4Z. An aqueous solution of K2Z was converted to Li2Z using the cation-exchange resin Purolite UCW 9126. The ion-exchange column was prepared using a 10 wt % aqueous solution of ACS reagent-grade LiCl, the specifications for which list the maximum amounts of K+ and Na+ as 0.01 and 0.02 mol %, respectively. The completeness of Li+/K+ cation exchange was determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES; see below). After a single pass through the cation-exchange column, the lithium salt contained 0.06 mol % K+, 0.27 mol % Na+, and 99.57 mol % Li+. Small amounts of Cu2+ (0.02 mol %) and Mg2+ (0.05 mol %) were also detected. The specifications for ACS-reagent-grade LiCl do not list these metal ions. All other metal ions that were measured by ICP-AES were less than 0.01 mol % (these were Ag, Al, As, Au, B, Ba, Be, Ca, Cd, Ce, Co, Cr, Cu, Fe, K, Li, Mg, Mn, Mo, Na, Ni, P, Pb, Pd, Pt, S, Sb, Se, Si, Sn, Sr, Te, Ti, Tl, U, V, W, Zn, and Zr). An aqueous solution of the lithium salt was passed through a freshly lithium-regenerated column and reduced the amount of K+ to less than 0.01 mol %. Anhydrous Li2(B12F12), which is hygroscopic when exposed to air, was isolated from the purified aqueous solution by rotary evaporation, followed by heating at 175−180 °C under vacuum for 24 h. The range of yields of anhydrous Li2Z from several syntheses was 85−90% based on the K2Z starting material. Recrystallization from water afforded single crystals of Li2(H2O)4Z suitable for X-ray diffraction (XRD) and TGA. It is likely that the use of a higher-purity grade of LiCl would result in even less contamination with Na+, K+, and other metal ions. LiK(H2O)4Z. A 50:50 mol % aqueous solution of Li2Z and K2Z was allowed to slowly evaporate, yielding single crystals of LiK(H2O)4Z suitable for XRD and TGA. Optimization of the amount of crystalline LiK(H2O)4Z isolated was not attempted, and therefore a yield was not determined. Na2(H2O)3Z. An aqueous solution of Na2Z was stored over a large amount of solid Ca(NO3)2 in a desiccator at 22(1) °C [the vapor pressure of H2O over a saturated solution of Ca(NO3)2 at 22(1) °C is 10 Torr]. Slow evaporation of the solution resulted in single crystals of Na2(H2O)3Z suitable for XRD and TGA. Optimization of the amount of crystalline Na2(H2O)3Z isolated was not attempted, and therefore a yield was not determined, in part because slight changes in this procedure sometimes resulted in the crystallization of Na2(H2O)4Z instead of Na2(H2O)3Z. Rb2(H2O)2Z. An aqueous mixture of RbCl and Ag2(CH3CN)4Z was filtered to remove AgCl. Slow evaporation of the filtrate resulted in single crystals of Rb2(H2O)2Z suitable for XRD and TGA. Optimization of the amount of crystalline Rb2(H2O)2Z isolated was not attempted, and therefore a yield was not determined. TGA. Samples for TGA (Pt sample pans; 10−15 mg of finely ground microcrystalline powders) were analyzed using TA Instruments series 2950 or TGA Q500 instrumentation. The temperatures for isothermal TGA experiments, which ranged from 23 to 32 °C, were held to within 0.01 °C for several minutes before a change in the carrier gas was made. Dry He, dry N2 or either gas bubbled through H2O, D2O, or a saturated aqueous solution of NaCl was used as the carrier gas depending on the experiment. When the carrier gas was bubbled through H2O, D2O, or saturated NaCl(aq) at 21(1) °C, the vapor pressure of H2O or D2O in the sample chamber was 18(1) Torr of H2O(g), 16(1) Torr of D2O(g), or 14(1) Torr of H2O(g), respectively. Whenever the carrier gas was switched [e.g., from dry gas to a gas containing H2O(g) or from gas containing H2O(g) to gas containing D2O(g) or vice versa], ca. 0.5 min elapsed before the composition of the carrier gas in the sample chamber became constant, as monitored in several control experiments by recording the mass spectrum of the carrier gas exiting the sample chamber (see Figures S1 and S2, which are reproduced from ref 1). The carrier gas flow rate was 60 mL min−1. ICP-AES. Metal analyses were performed using PerkinElmer model 7300 DV ICP-OES and ICP-AES instruments. Elements analyzed were Ag, Al, As, Au, B, Ba, Be, Ca, Cd, Ce, Co, Cr, Cu, Fe, K, Li, Mg, Mn, Mo, Na, Ni, P, Pb, Pd, Pt, S, Sb, Se, Si, Sn, Sr, Te, Ti, Tl, U, V,

In addition, the reversible transformations K2(H2O)2Z + 2D2O(g) ⇌ K2(D2O)2Z + 2H2O(g) were >90% complete in only 120 min at 25 °C when the partial pressure of water in the He gas stream was changed from 21 Torr of H2O(g) to 18 Torr of D2O(g) and back to 21 Torr of H2O(g). The unusually rapid and repeatable room-temperature hydration/dehydration cycles and H2O/D2O exchange cycles in crystalline solids that are not microporous, viz., K2Z and K2(H2O)2Z,1 were termed latent porosity, which we defined as the rapid creation of space in a nonporous crystal lattice in response to the presence of reactive gases at T = 25 ± 5 °C.1 The structures of K2Z,2 K2(H2O)2Z,1 K2(H2O)4Z,1 and Cs2(H2O)Z2 were reported in 2010. The structures of Na2Z7 and Na2(H2O)4Z7 were reported in 2016. In this work, we report the dehydration and rehydration behavior of M2(H2O)nZ compounds of all five alkali-metal ions (n = 0−4) by isothermal and/or nonisothermal thermogravimetric analysis (TGA). This includes new results for K2(H2O)2Z ↔ K2Z dehydration/rehydration cycles, the demonstration that Rb2Z, Cs2Z, and Na2(H2O)2Z also exhibit latent porosity, the temperature dependence of the rates of hydration and dehydration, and the rapid exchange of H2O and D2O molecules when crystalline Rb2(H2O)2Z is exposed to 16 Torr of D2O(g) or when Rb2(D2O)2Z is exposed to 18 Torr of H2O(g). We also report the single-crystal X-ray diffraction (SC-XRD) structures of Li 2 (H 2 O) 4 Z, LiK(H 2 O) 4 Z, Na2(H2O)3Z, and Rb2(H2O)2Z, the powder X-ray diffraction (PXRD)-determined unit cell parameters for Rb2Z and Rb2(H2O)2Z at room temperature, and the density functional theory (DFT)-optimized structures of K2Z, K2(H2O)2Z, Rb2Z, Rb2(H2O)2Z, Cs2Z, and Cs2(H2O)Z. The structures support possible explanations for the unusually rapid solid-state dehydration/rehydration and H2O/D2O exchange reactions at room temperature. In addition, because Li2Z has applications as an electrolyte for secondary batteries and other electrochemical devices,35−39 we also report a method to isolate it with highpurity and minimal residual H2O.

2. EXPERIMENTAL METHODS Reagents and General Procedures. Anhydrous compounds were prepared using standard airless-ware glassware and a Schlenkstyle vacuum line and stored in a nitrogen-filled glovebox.40 Potassium dodecafluoro-closo-dodecaborate(2−), K2Z (Z2− = B12F122−), was synthesized by the direct fluorination of K2B12H12 with 80:20 N2/F2 in CH3CN at 0 °C and purified as previously described.5,41,42 [Caution! The original purif ication procedure41,42 involving H2O2 is not recommended because of the potential isolation of explosive K2(H2O2)2−x(H2O)xZ. An alternate, safer procedure5 is recommended.] The substitution of F atoms for H atoms was monitored periodically using negative-ion electrospray-ionization mass spectrometry (NI-ESIMS), as previously described.41 The final degree of F/H metathesis, determined by NI-ESI-MS and 19F{11B} and 11B{19F} NMR spectroscopy as previously described,41 was found to be 99.5+%. Recrystallization from water removed a trace amount of BF4− that was present. It is important to remove BF4− contamination from the potassium salt because it was not possible to efficiently remove BF4− from either Li2Z or Na2Z by recrystallization. The hydrated salts Na2(H2O)4Z7 and Cs2(H2O)Z2 and the solvated salt Ag2(CH3CN)4Z43,44 were prepared as previouly described. Distilled water was deionized with a Barnstead Nanopure system. The deionized distilled water (dd-H2O) had a resistivity greater than or equal to 18 MΩ (all samples of H2O used in this work correspond to dd-H2O prepared in this way). The following reagents were obtained from the indicated suppliers and used as received: deuterium oxide (D2O, Cambridge Isotopes, 99.9% D); LiCl (Mallinckrodt, ACS reagent grade, lot G21621); RbCl (K&K, 99%). [The mention of all commercial suppliers in this paper is for clarity. 12024

DOI: 10.1021/acs.inorgchem.7b02081 Inorg. Chem. 2017, 56, 12023−12041

Article

Inorganic Chemistry

Table 1. Crystal Data and Final Refinement Parameters for the Single-Crystal X-ray Structures Reported in This Worka compound

Li2(H2O)4(B12F12)

LiK(H2O)4(B12F12)

Na2(H2O)3(B12F12)

Rb2(H2O)2(B12F12)

formula fw, g mol−1 cryst syst space group, Z a, Å b, Å c, Å α, deg β, deg γ, deg V, Å3 ρcalc g cm−3 T, K R(F) [I > 2σ(I)]b wR(F2) (all data)b GOF

B12F12H8Li2O4 443.66 orthorhombic Ibam, 4 10.3005(11) 10.3100(13) 13.5337(13) 90 90 90 1437.3(3) 2.050 120(2) 0.0295 0.1488 1.343

B12F12H8KLiO4 475.82 tetragonal P42/ncm, 4 10.5921(5) 10.5921(5) 14.0638(9) 90 90 90 1577.9(2) 2.003 120(2) 0.0463 0.1694 1.205

B12F12H6Na2O3 457.75 orthorhombic P212121, 4 10.2974(6) 10.3442(6) 14.1466(9) 90 90 90 1506.9(2) 2.018 130(2) 0.0339 0.0910 1.067

B12F12H4O2Rb2 564.69 monoclinic P21/c, 2 7.8511(3) 10.2749(4) 9.9917(4) 90 108.775(1) 90 763.13(5)c 2.457 120(2) 0.0184 0.0422 1.051

The radiation used was 0.71073 Å X-rays. bR(F) = ∑||Fo| − |Fc||/∑|Fo|; wR(F2) = (∑[w(Fo2 − Fc2)2]/∑[w(Fo2)2])1/2. cThe unit cell volume at 297 K, determined by PXRD, is 783.2(1) Å3, 2.6% larger.

a

W, Zn, and Zr. Samples were diluted in a mixture of metal-free acids [1% (v/v) HCl and 5% (v/v) HNO3] to a known concentration, and the reported values were an average of three replicate measurements. Y was used as an internal standard, and 15 interelement standards were used to correct for easily ionizable elements. SC-XRD. Data sets were collected using a Bruker Kappa APEX II CCD diffractometer, and structures were solved using standard Bruker X-ray software. Table 1 lists relevant data collection and refinement results for the four new SC-XRD structures. Tables 2 and 3 lists relevant interatomic distances and angles for the new structures and, for comparison, the previously published structures of Na2Z,7 Na2(H2O)4Z,7 Ag2(H2O)4Z,45 K2Z,2 K2(H2O)2Z,1 K2(H2O)4Z,1 and Cs2(H2O)Z,2 including distances involving the B12 centroids, hereinafter denoted with the symbol ⊙. Table S-1 lists the bond-valence parameters used in this work.46 PXRD. Powder patterns for samples of Rb2Z and Rb2(H2O)2Z in sealed quartz capillaries were collected at 24 °C using a Rigaku Ultima III powder X-ray diffractometer with Cu Kα radiation. XRD patterns and Le Bail refinement fits to the patterns for polycrystalline samples of anhydrous Rb2Z and for Rb2(H2O)2Z are shown in Figure S3. Indexing of the PXRD pattern for Rb2Z revealed that it is isomorphous with K2Z (space group C2/c). The lattice parameters determined from the Le Bail refinement for Rb2Z are a = 8.418(1) Å, b = 14.704(2) Å, c = 11.610(2) Å, β = 93.054(7)°, Z = 4, and V = 1435.1(5) Å3. Indexing of the PXRD pattern for Rb2(H2O)2Z was consistent with the P21/c SC-XRD structure determined at 120 K. The lattice parameters determined from the Le Bail refinement for Rb2(H2O)2Z are a = 7.9109(4) Å, b = 10.3972(6) Å, c = 10.0599(5) Å, β = 108.831(2)°, Z = 2, and V = 783.2(1) Å3. DFT Calculations. First-principles calculations were performed for K2Z, K2(H2O)2Z, Rb2Z, Rb2(H2O)2Z, Cs2Z, and Cs2(H2O)Z within the plane-wave implementation of the generalized gradient approximation to DFT using a Vanderbilt-type ultrasoft potential with Perdew− Burke−Ernzerhof (PBE) exchange correlation.47 A cutoff energy of 544 eV and a 2 × 2 × 2 k-point mesh (generated using the Monkhorst− Pack scheme) were used and found to be enough for the total energy to converge to within 0.01 meV atom−1. Unit cell parameters, fractional coordinates, bond distances, and bond valences for DFT K2Z and K2(H2O)2Z, bond distances and bond valences for SC-XRD K2Z and K2(H2O)2Z, and selected distances and volumes for the DFT and SC-XRD structures of K2Z and K2(H2O)2Z are listed in Tables S2−S8. Unit cell parameters, fractional coordinates, bond distances, and bond valences for Rb2Z, Rb2(H2O)2Z, Cs2Z, and Cs2(H2O)Z and bond distances and bond valences for SC-XRD Rb2(H2O)2Z and Cs2(H2O)Z are listed in Tables S9−S18. Selected distances and volumes are listed in Table 3.

3. RESULTS AND DISCUSSION I. Attempts to Crystallize Mixed-Cation Salts. Crystalline salts containing both K+ and Cs+ are known, including KCs(ClO 4 ) 2 , 4 8 KCs(Pd(NO 3 ) 4 )·0.5H 2 O, 49 and KCs 2 (Bi(SCN)6).50 However, an aqueous solution containing K+, Cs+, and Z2− deposited crystals of the sparingly soluble salt Cs2(H2O)Z, not KCs(H2O)nZ (Z2− = B12F122−). This may be because the structures of Cs2(H2O)Z2 and both K2(H2O)2Z1 and K2(H2O)4Z1 are substantially different. Nevertheless, an aqueous solution containing Li+, K+, and Z2− deposited crystals of the hydrated mixed-cation salt LiK(H2O)4Z in spite of the substantial differences between the structures of Li2(H2O)4Z, LiK(H2O)4Z, and K2(H2O)4Z, as discussed below. Some other mixed-cation salts containing Li+ and a different alkali-metal ion are LiCsTiF6,51 LiK(BH4)2,52 LiK(N(CN)2)2,53 LiRb(N(CN)2)2,53 and LiK2(SCN)3.54 Two salts containing Na+ and K+ are NaKSiF655 and Na2K2P2O6·8H2O.56 There are a number of other examples with octahedral MF6n− and related fluoroanions.57 No mixedcation salts of Z2− other than LiK(H2O)4Z have been reported. II. New M2(H2O)nZ Structures. II.A. General Comments. Relevant interatomic distances, B12 centroid···centroid (⊙···⊙) distances, sums of the M−O and M−F bond valences (∑bv values), and formula unit volumes are listed in Table 2 for the SC-XRD structures of Li2(H2O)4Z, Na2(H2O)3Z, Na2(H2O)4Z,7 LiK(H2O)4Z, K2(H2O)4Z,1 and Ag2(H2O)4Z45 and in Table 3 for the SC-XRD structures of K2(H2O)2Z,1 Rb2(H2O)2Z, and Cs2(H2O)Z2 and the DFT-optimized structures of Rb2Z, Rb2(H2O)2Z, Cs2Z, and Cs2(H2O)Z. As expected, distances and angles within the icosahedral Z2− anions in all of the structures are essentially the same and will not be discussed further. A recent review lists metal-ion and NH4+ salts of B12X122− anions that were known at the end of 2015 (X = H, F, Cl, Br, I).58 In addition to the new SC-XRD structures of Z2− reported in this paper, the SC-XRD structures of K2(SO2)6Z, Ag2(SO2)6Z, Ag2(H2O)4Z, Ag2(CH3CN)4,5,8Z, Ag2(CH2Cl2)4Z, Ag 2 (C 6 H 5 CH 3 ) 6 Z, Na 2 Z, Na 2 (H 2 O) 4 Z, Na 2 (B 12 Cl 12 ), Na2(H2O)6(B12Cl12), and PPh4+ salts of B12X11(OR)2− anions (X = Cl, Br; R = H, n-Pr, n-Oc) were also reported since 2015.7,43,45,59 II.B. Li2(H2O)4Z. The structure of this compound, shown in Figures 1 and S4 and S5, is unlike the structures of any of the other hydrated alkali-metal and silver salts of Z2−. The B12 12025

DOI: 10.1021/acs.inorgchem.7b02081 Inorg. Chem. 2017, 56, 12023−12041

offset hexagonalf 7.032−8.971 3.516(1) (Li···K)

offset hexagonalf 6.888−8.798 3.558(6)−3.609(6)

1.01−1.14 0.28−0.34 0.73−0.79 5.037−6.303

1.945(3) (×4) (Li); 2.856(3) (×4) (K) 1.10 (Li); 1.27 (K) 0.70 (K) 1.10 (Li); 0.56 (K) 5.843 × 4 (Li); 5.399 (×4) (K)

2.368(9)−2.473(9)

2.632(2) (×4) (K)

LiO4; KO4F4

LiK(H2O)4Z

distorted CsCl 6.915−9.126 4.3188(4)

1.09 (K1); 1.03 (K2) 0.55 (K1); 0.49 (K2) 0.54 (K1); 0.55 (K2) 5.192−5.930

2.6685(7)−2.8306(8)

2.6501(6)−3.2843(7)

KO3F5

K2(H2O)4Zc

0.87 (Ag1); 0.90 (Ag2) 0.12 (Ag1); 0.16 (Ag2) 0.75 (Ag1); 0.74 (Ag2) 5.159−6.041 (Ag1); 5.117−5.959 (Ag2) offset hexagonalf 7.003−8.429 3.540(2)

2.371(4)−2.524(4)

2.734(4)−3.021(4)

AgO4F2

Ag2(H2O)4Zd

12026

4.51, 4.79e

P21/c 381.6 distorted CsCl 7.17, 7.85, 9.37 6.96

4.40, 4.80e

space group FUV (Å3)d anion packing ⊙···⊙ distance between anion layers closest M···M 5.59f

4.89−5.88 (Rb1); 5.64−5.90 (Rb2) C2/c 370.7 hexagonal 8.51, 8.55 5.87

1.11 (Rb1); 0.68 (Rb2) 1.11, 0.68

RbF8 (Rb1); RbF8 (Rb2) 2.81−3.00 (Rb1); 2.84−3.44 (Rb2)

Rb2Z DFT

b

c

4.93, 5.49g d

DFT

4.97, 5.60g

CsF11 (Cs1); CsO2F7 (Cs2) 3.13−3.42 (Cs1); 3.15−3.51 (Cs2) 3.23, 3.35 (Cs2) 0.94 (Cs1); 0.79 (Cs2) 0.94, 0.60 0.19 (24%) 5.10−6.08 (Cs1); 5.24−8.29 (Cs2) P212121 391.8 hexagonal 8.15−10.31 5.21 ± 0.07

Cs2(H2O)Z CsF11 (Cs1); CsO2F7 (Cs2) 3.05−3.44 (Cs1); 3.07−3.47 (Cs2) 3.15, 3.30 (Cs2) 1.03 (Cs1); 0.94 (Cs2) 1.03, 0.71 0.23 (24%) 5.03−6.01 (Cs1); 5.16−8.14 (Cs2) P212121 376.3 hexagonal 8.08−10.16 5.09 ± 0.07

SC-XRDc

e

5.12, 5.56g

5.04−5.97 (Cs1); 5.15−7.98 (Cs2) P212121 386.6 hexagonal 8.02−10.10 5.28 ± 0.05

0.87 (Cs1); 0.65 (Cs2) 0.87, 0.65

CsF10 (Cs1); CsF8 (Cs2) 3.15−3.36 (Cs1); 3.14−3.61 (Cs2)

Cs2Z DFT

4.25

P21/c 367.9 distorted CsCl 7.28, 9.23 7.09

2.77, 2.77 1.01 0.65 0.36 (36%) 4.69−5.66

KO2F6 2.59−3.16

K2(H2O)2Zb SC-XRD

5.42h

4.72−5.67 (K1); 5.45−5.65 (K2) C2/c 332.0 hexagonal 8.21, 8.24 5.67

1.16 (K1); 0.73 (K2) 1.16, 0.73

KF8 (K1); KF8 (K2) 2.65−3.41 (K1); 2.62−3.32 (K2)

K2Zc SC-XRD

Z = B12F12 . All results are from this work unless otherwise indicated. Reference 1. Reference 2. FUV = formula unit volume. The shorter distance is between Rb atoms in different Rb2(H2O)2 rhombs; the longer distance is between Rb atoms in the same rhomb. fThis is the shortest Rb1···Rb2 distance. gThe shorter distance is between Cs1 and Cs2 atoms; the longer distance is between two Cs2 atoms. hThis is the shortest K1···K2 distance.

2−

P21/c 399.5 distorted 7.30, 7.93, 9.49 7.09

2.86, 2.92 1.04 0.67 0.37 (36%) 4.97−5.91

M−O ∑bv(M−F/O) ∑bv(M−F) ∑bv(M−O) M···⊙

a 2−

2.91, 2.94 0.87 0.54 0.33 (38%) 5.08−5.94

RbO2F8 2.91−3.24

DFT

RbO2F8 2.94−3.35

Rb2(H2O)2Z

SC-XRD

M coordination sphere M−F(B)

compound

Table 3. Selected Interatomic Distances (Å) and Bond Valence Valuesa

Z = B12F122−. All results are from this work unless otherwise indicated. bReference 7. cReference 1. dReference 45. eNa1 has a trans-NaO2F4 coordination sphere; Na2 has a cis-NaO2F4 coordination sphere. fADAD... packing (see the Supporting Information for more details).

tetragonal 6.767, 7.287 (×2) 5.150(1)

anion packing ⊙···⊙ closest M···M

a 2−

1.00 0.52 0.48 4.972

∑bv(M−F/O) ∑bv(M−F) ∑bv(M−O) M···⊙

M−O

distorted CsCl 7.185−7.475 4.229(1)

2.267(2)−2.293(2) (Na1); 2.292(2)−2.636(2) (Na2) 2.402(2), 2.434(2) (Na1); 2.387(2), 2.416(2) (Na2) 1.16 (Na1); 1.04 (Na2) 0.78 (Na1); 0.64 (Na2) 0.38 (Na1); 0.40 (Na2) 5.119 (Na1); 4.645 (Na2)

2.109(1) (×2), 2.118(1) (×2) 1.995(1) (×2) (K)

2.313(8)−2.381(7)

NaO4F2

cis- and trans-NaO2F4e

LiO2F4

M coordination sphere M−F(B)

Na2(H2O)4Zb

Li2(H2O)4Z

compound

Na2(H2O)3Z

Table 2. Selected Interatomic Distances (Å) and Bond Valence Values for Single-Crystal X-ray Structuresa

Inorganic Chemistry Article

DOI: 10.1021/acs.inorgchem.7b02081 Inorg. Chem. 2017, 56, 12023−12041

Article

Inorganic Chemistry

of 359 Å3. The anion parallelpiped for Li2(H2O)4(B12H12) has a volume of 340 Å3, only 6% smaller (see Figure S6). Nevertheless, one determining factor may be the relative strengths of the Li−F and Li−H bonds. Another factor may be the absence of O(H)···O hydrogen bonding in Li2(H2O)4(B12H12) and the presence of two O(H)···O hydrogen bonds per H2O molecule in Li2(H2O)4Z, which also exhibits additional, albeit weaker, O(H)···F hydrogen bonds. A lithium salt hydrate with a square-pyramidal Li(H2O)F4 coordination sphere and discrete F2(H2O)Li(μ-F2)Li(H2O)F2 dimeric moieties is Li2Mg(H2O)4(ZrF6)2 [Li−O = 2.016(3) Å; ∑bv(Li−X) = 0.97].63 There are a number of lithium salt hydrates of other fluoroanions that, like Li2(H2O)4Z, have Li(H2O)2 moieties and either cis- or trans-LiO2F4 coordination spheres. These are Li(H 2O)(AsF 6 ),64 Li(H 2 O)(BF 4 ),65 Li2(H2O)2(TiF6),66 and Li2(H2O)2(SnF6).66 Table S21 lists Li−O and Li−F distances and bond valence sums [∑bv(Li−X)] and Li···Li distances for these four structures and for Li2(H2O)4Z. The most striking difference is that the H2O/Li ratio is 2:1 in Li2(H2O)4Z and 1:1 in the other compounds. This can be attributed to the much larger size of Z2− relative to BF4−, AsF6−, TiF62−, and SnF62−. For example, the Li−O and Li−F bond valence values for the trans-LiO2F4 coordination spheres in Li2(H2O)4Z and Li2(H2O)2(SnF6) add up to 1.00 and 0.97, respectively. The 1:1 H2O/Li stoichiometry in Li2(H2O)2(SnF6) is due to the formation of [−Li(μ-H2O)Li (μ-H2O)−]∞ chains, with Li···Li distances of 3.05 Å. The Sn−F, Sn···Sn, and Li···Sn distances are 1.97, 4.73, and 3.71 Å, respectively. In contrast, the ⊙···F, ⊙···⊙, and Li···⊙ distances in Li2(H2O)4Z are 3.07, 6.77, and 4.97 Å, respectively. The anion lattice in Li2(H2O)4Z is presumably too big to allow [−Li (μ-H2O)nLi(μ-H2O)n−]∞ chains with Li···Li distances of only 3.1 Å to form while maintaining optimal Li−F distances. Note that the other three compounds in Table S21 also have [−Li(μ-H2O)Li(μ-H2O)−]∞ chains. Larger metal ions such as Na+ and Ag+ can form such chains in a large Z2− lattice: the Na···Na and Ag···Ag distances in Na2(H2O)4Z7 and Ag2(H2O)4Z45 are 3.56−3.61 and 3.54 Å, respectively, and both compounds have 2:1 H 2 O/M stoichiometries and [−M(μ-H2O)2M(μ-H2O)2−]∞ chains. The same is true for the mixed-metal hydrate LiK(H2O)4Z, which we will discuss next. II.C. LiK(H2O)4Z. The structure of this compound is shown in Figure 2. It consists of partially offset hexagonal layers of Z2− anions with parallel [−Li(μ-H2O)2K(μ-H2O)2−]∞ chains between them aligned with the ADAD··· offset direction (see below). The chains consist of tetrahedral LiO4 and compressed square-antiprismatic KO4F4 coordination spheres that are bridged through pairs of H2O ligands (the KO4F4 coordination sphere is shown in Figure S7). Bond distances and bond valences are listed in Table S22. As mentioned in the previous paragraph, the structures of LiK(H2O)4Z, Na2(H2O)4Z, and Ag2(H2O)4Z are virtually the same (except that both metal ions in the latter two compounds have MO4F2 coordination spheres), with parallel [−M(μ-H2O)2M(μ-H2O)2−]∞ chains between offset hexagonal layers of Z2− anions [i.e., ADAD··· anion packing,67 as shown in Figure S8 for LiK(H2O)4Z and several related structures]. The formula unit volumes are 395 Å3 (LiK), 394 Å3 (Na), and 400 Å3 (Ag). The M···M distances are also virtually the same, 3.516(1) Å (LiK), 3.56−3.61 Å (Na), and 3.54 Å (Ag). The [−M(μ-H2O)2M(μ-H2O)2−]∞ chains in LiK(H2O)4Z and Na2(H2O)4Z are compared in Figure 3. Not only is the structure of LiK(H2O)4Z significantly

Figure 1. Structure of Li2(H2O)4(B12F12) (50% probability ellipsoids except for H atoms). Only one of the eight Z2− anions at the corners of the tetragonal unit cell is shown, and only one Li+ ion is shown with its complement of four F atoms. The trans-LiO2F4 coordination sphere is shown in Figure S4. The unique Li−O distance is 1.995(1) Å. The two unique pairs of Li−F distances are 2.109(1) and 2.118(1) Å. The (H2O)4 cluster in the center of the tetragonal cell forms an O4 skew-quadrilateral with pairs of O(H)···O distances of 2.778(1) and 2.785(1) Å and a dihedral angle of 71.7°. Another view of the skew-quadrilateral is shown in Figure S5.

centroids (⊙) form a nearly perfect tetragonal array, with ⊙···⊙ distances of 6.767, 7.287, and 7.287 Å and with ⊙···⊙···⊙ angles within 0.1° of 90°. The Li+ ions are centered on the four 6.767 × 7.287 Å faces. Each Li+ ion is coordinated to four F atoms, one from each of four Z2− anions, with pairs of Li−F distances of 2.109(1) and 2.118(1) Å, and to two H2O molecules arranged so that (i) the symmetry-related Li−O distances are 1.995(1) Å, (ii) the O−Li−O angle is 180°, and (iii) the H2O molecules do not bridge two Li+ ions. Instead, each tetragonal array of B12 ⊙’s contains four hydrogen-bonded H2O molecules that form an O4 skew-quadrilateral, with pairs of 2.778(1) and 2.785(1) Å O(H)···O distances and a dihedral angle of 71.7° [cf. crystalline H2O-Ih, with an average O(H)···O distance of 2.76 Å]. Clusters of four H2O molecules in solidstate structures are generally flat, nearly square arrays,60 which is also the theoretically lowest-energy structure for an isolated (H2O)4 cluster.61 The H atoms that are not involved in O(H)···O hydrogen bonding are weakly hydrogen-bonded to F atoms of the Z2− anion [there are four O(H)···F distances that range from 2.88 to 2.97 Å]. The Li−O and Li−F bond distances and bond valences are listed in Table S19. A related structure with a cubic array of Z2− anions and ⊙···⊙ distances of 7.242 Å is K2(HF)3Z [in that case, K2(μ-HF)32+ cations are centered in each cube of anions].6 The structure of Li2(H2O)4(B12H12), with the same 2:4:1 stoichiometry as Li2(H2O)4Z, has been determined by PXRD.62 In contrast to the structure of Li2(H2O)4Z, there are discrete (H2O)Li(μ-H2O)2Li(H2O) dinuclear complexes, as shown in Figure S6, in which each Li+ ion is coordinated to two bridging H2O ligands, one terminal H2O ligand, and a B12H122− anion in tridentate fashion (i.e., the Li+ ions are six-coordinate, with three Li−O bonds and three Li−H bonds; see Table S20 for a list of bond distances and bond valences).62 It is not clear why the cations and H2O molecules are organized differently in the two structures, especially because the anion lattices are quite similar in size and shape. The anion parallelpiped for Li2(H2O)4Z shown in Figure 1, with the ⊙···⊙ distances and ⊙···⊙···⊙ angles listed in the previous paragraph, has a volume 12027

DOI: 10.1021/acs.inorgchem.7b02081 Inorg. Chem. 2017, 56, 12023−12041

Article

Inorganic Chemistry

Figure 2. Structure of LiK(H2O)4(B12F12). The upper drawing, shown with 50% probability ellipsoids except for H atoms, shows a segment of the [−Li(μ-H2O)2K(μ-H2O)2−]∞ chains of tetrahedral LiO4 and compressed square-antiprismatic KO4F4 coordination spheres. The lower drawing shows two parallel [−Li(μ-H2O)2K(μ-H2O)2−]∞ chains between two offset hexagonal layers of Z2− anions (H, B, and F atoms omitted for clarity; the anions are represented by their B12 centroids).

different from the structure of Li2(H2O)4Z, it is also significantly different from the structure of K2(H2O)4Z, which is also shown for comparison in Figure 3. It is of interest that the sum of the four- and eight-coordinate effective ionic radii of Li+ and K+ is 2.10 Å (i.e., 0.59 Å + 1.51 Å = 2.10 Å) and twice the sixcoordinate effective ionic radius of Na+ is 2.04 Å (i.e., 2 × 1.02 Å = 2.04 Å), virtually the same.68 The four symmetry-related Li−O distances in LiK(H2O)4Z are 1.945(3) Å, and the sum of the Li−O bond valences is 1.10. A compound with somewhat similar Li(H2O)4 coordination spheres is Li2(H2O)7(B12H12), shown in Figure S10, in which one of the H2O molecules bridges two Li(H2O)3 moieties. The Li−O distances are 1.929, 1.942 × 2, and 2.010 Å (s.u.’s were not reported),69,70 and the sum of the Li−O bond valences is 1.07. Compounds with discrete Li(H2O)4+ cations include Li(H 2 O) 4 (Li(C 7 H 5 O 2 ) 2 ), 71 Li(H 2 O) 4 (B(OH) 4 )·2H 2 O, 72 Li(H2O)4(OC(CF3)3),73 and Li(H2O)4(Al(OC(CF3)3)4).73 An interesting structure is Li(H2O)3(AsF6),74 which is shown in Figure S10. It contains [−Li(μ-H2O)3Li(μ-H2O)3−]∞ chains with rare six-coordinate Li(H2O)6 coordination spheres [Li−O = 2.120 Å × 3 and 2.119 Å × 3; ∑bv(Li−O) = 1.03; Li···Li = 2.746 Å]. II.D. Na2(H2O)3Z. The structure of this compound, shown in Figures 3 and 4, is unique among Z2− salt hydrates characterized to date in that one of the H2O molecules is not bonded to a metal ion; it is only hydrogen-bonded to the other H2O molecules. The structures of Li2(H2O)4Z (Figure 1) and K2(H2O)4Z1 (Figure 3) also contain H2O molecules that participate in O(H)···O hydrogen bonds, but in those cases, all of the H2O molecules are bonded to the metal ions. There are alternating cis- and trans-NaO2F4 coordination spheres in Na2(H2O)3Z that are linked to form [−Na(μ-H2O)Na (μ-H2O)−]∞ chains (the individual Na−O and Na−F bond distances and bond valences are listed in Table S23). The lattice H2O molecule acts as a bifurcated hydrogen-bond acceptor for the pair of H2O molecules in each cis-NaO2F4 moiety. This forms a four-atom NaO3 array that is planar to

Figure 3. Comparison of 50% probability ellipsoid drawings (except for H atoms) of the structures of Na2(H2O)3Z (top; this work; Z2− = B12F122−), Na2(H2O)4Z (second from top; ref 7), LiK(H2O)4Z (second from bottom; this work), and K2(H2O)4Z (bottom; ref 1). Note the similarities between the structures of LiK(H2O)4Z and Na2(H2O)4Z and the differences between the structures of LiK(H2O)4Z and K2(H2O)4Z.

within ±0.01 Å. It is sensible that the coordinated H2O molecules are the hydrogen-bond donors because, by virtue of their coordination to two Na+ ions, they should have more acidic H atoms than a lattice H2O molecule. An interesting feature of the structure of Na2(H2O)3Z is that the anions are not arranged in flat close-packed layers because they are in Na2Z,7 Na2(H2O)4Z,7 LiK(H2O)4Z, K2Z,2 K2(H2O)4Z,1 and Ag2(H2O)4Z45 and several silver(I) solvate salts including Ag 2 (CH 3 CN) 4 Z and Ag2(CH2Cl2)4Z.43 (Some of these rigorously coplanar or nearly coplanar layers of B12 centroids are shown in Figure S8.) Instead, the undulating snakelike [−Na(μ-H2O)Na(μ-H2O)−]∞ 12028

DOI: 10.1021/acs.inorgchem.7b02081 Inorg. Chem. 2017, 56, 12023−12041

Article

Inorganic Chemistry

Figure 4. Part of the structure of Na2(H2O)3(B12F12) (50% probability ellipsoids except for H atoms), showing the alternating cis- and transNaO2F4 coordinating units that are linked to [−Na(μ-H2O)Na (μ-H2O)−]∞ chains. The O(H)···O hydrogen-bond distances are 2.710(4) and 2.799(4) Å.

chains in Na2(H2O)3Z appear to cause the anions to pack in corrugated layers, as shown in Figure S11. The [−Na(μ-H2O)Na(μ-H2O)−]∞ chains in Na2(H2O)3Z and the [−Na(μ-H2O)2Na(μ-H2O)2−]∞ chains in Na2(H2O)4Z are shown next to one another in Figure 3. It can be seen that three B−F bonds belonging to a Z2− triangular face tightly bridge neighboring Na+ ions along the chain in Na2(H2O)3Z. The Z2− anions do not bridge neighboring Na+ ions along the [−Na(μ-H2O)2Na(μ-H2O)2−]∞ chains in Na2(H2O)4Z. II.E. Rb2(H2O)2Z. The SC-XRD structures of Rb2(H2O)2Z, shown in Figure 5, and of K2(H2O)2Z1 are very similar (drawings of their structures are shown side-by-side in Figure S12). Their P21/c unit cells have similar volumes (763 vs 736 Å3, respectively), their average ⊙···⊙ distances are 7.39 and 7.28 Å3, respectively, and the Rb···Rb and K···K distances within the M2(μ-H2O)2 rhombs are 4.797(1) and 4.2529(3) Å, respectively. The bond valence sums ∑bv(M−O) and ∑bv(M−F) are 0.37 and 0.67, respectively, for Rb2(H2O)2Z and 0.35 and 0.65, respectively, for K2(H2O)2Z. Two differences are (i) RbO2F8 versus KO2F6 coordination spheres and (ii) the shortest Rb···Rb distance is 4.400(1) Å, shorter by ca. 0.4 Å than the distance between Rb+ ions in the same Rb2(μ-H2O)2 rhomb. In contrast, the shortest distance between K+ ions that are not in the same K2(μ-H2O)2 rhomb is 5.0750(3) Å, more than 0.8 Å longer that the distance between K+ ions in the same K2(μ-H2O)2 rhomb. A comparison of the SC-XRD structure of Rb2(H2O)2Z with other rubidium salt hydrates will be discussed in section III.E. To validate the DFT code that we used to calculate the structures of hydrated and anhydrous Rb+ and Cs+ salts of Z2−, we first compared the SC-XRD and DFT-optimized structures of K2Z and K2(H2O)2Z. The results are shown in Tables S2−S4. The pairs of structures are virtually the same except that the DFT distances are greater, as is commonly found when using a Vanderbilt-type ultrasoft potential with PBE exchange correlation (i.e., the interaction potentials are typically underestimated, resulting in larger lattice constants than those experimentally observed).75 The expansion of the K2Z formula unit volume upon hydration to K2(H2O)2Z is 10.8% (18.0 Å3 per H2O) for the 120/110 K SC-XRD structures1,2 and 11.9% (20.8 Å3 per H2O) for the DFT structures. Bond distances and bond valences for the SC-XRD and DFT structures of K2Z and K2(H2O)2Z are listed in Tables S5−S8. The bond valences for the two K−OH2 bonds are 36 and 37% of the total bond

Figure 5. Structure of Rb2(H2O)2(B12F12) (50% probability ellipsoids except for H atoms). The irregular RbO2F8 coordination sphere is shown for only one of the symmetry-related Rb+ ions. The Rb−O distances (Å) are shown. The Rb−F distances range from 2.909(1) to 3.237(1) Å. The Rb···Rb distance in the Rb2(μ-H2O)2 rhomb is 4.797(1) Å.

valences for the SC-XRD and DFT structures of K2(H2O)2Z, respectively. Unit cell parameters, fractional coordinates, bond distances, and bond valences for the DFT-optimized structures of Rb2Z and Rb2(H2O)2Z, and bond distances and bond valences for the SC-XRD structure of Rb2(H2O)2Z, are listed in Tables S9−S13. Important distances and formula unit volumes are listed in Table 3. The DFT-optimized structure of Rb2Z and the SC-XRD structure of K2Z2 are isomorphous (C2/c; Z = 4). They have a B82-Ni2In-like structure,76 with half of the metal ions in Oh holes and half of the metal ions in D3h holes in the expanded close-packed layers of Z2− anions. The DFT and SC-XRD structures of Rb2(H2O)2Z are also very similar (both have P21/c lattices with Z = 2). These two structures are compared in Figure S13 and Table 3. At the DFT level of theory, the Rb2Z formula unit volume expansion upon hydration to Rb2(H2O)2Z is 7.77% (14.4 Å3 per H2O). Experimental unit cell parameters for Rb2Z and Rb2(H2O)2Z were determined by PXRD at room temperature (see the Experimental Methods section). The expansion of the Rb2Z formula unit volume upon hydration to Rb2(H2O)2Z was found to be 9.14% (16.4 Å3 per H2O) at 24 °C. There is no doubt that the Rb2Z lattice must expand in order to absorb H2O. In addition, because the PXRD sample of Rb2(H2O)2Z was prepared from the PXRD sample of Rb2Z by hydration in the presence of water vapor [hereinafter H2O(g)], there is also no doubt that the transformation Rb2Z + 2H2O → Rb2(H2O)2Z is a crystal-to-crystal transformation, similar to what was previously shown for the K2Z + 2H2O → K2(H2O)2Z transformation.1 II.F. DFT Structures of Cs2Z and Cs2(H2O)Z. Interestingly, the DFT computations for Cs2Z strongly suggest that its structure is not isomorphous with the C2/c DFT structure of 12029

DOI: 10.1021/acs.inorgchem.7b02081 Inorg. Chem. 2017, 56, 12023−12041

Article

Inorganic Chemistry

III. Dehydration/Rehydration of Z2− Salt Hydrates. III.A. Kinetic versus Thermodynamic Factors. In 2010, we reported that microcrystalline K2Z was hydrated to K2(H2O)2Z in a matter of minutes at 25 °C in a continuous purge of He containing 21 Torr of H2O(g) and dehydrated at 25 °C in a continuous purge of dry He in less than 1 h.1 In addition, we determined that (i) the equilibrium vapor pressure of water [P(H2O)] for the K2(H2O)2Z ⇌ K2Z + 2H2O(g) equilibrium at 25 °C is 6.1(3) Torr and (ii) the enthalpy change (ΔH) for the forward reaction is 55.5 kJ (mol of H2O)−1.1 The ΔH values for the partial or complete dehydration of eight other potassium salt hydrates range from 56.0 to 70.2 kJ (mol of H2O)−1 [av. 60.7 kJ (mol of H2O)−1].77 Figure S14 is a plot of ΔH° versus P(H2O) for these eight compounds and for K2(H2O)2Z (the individual values77 are listed in Table S24). It is clear that the K2(H2O)2Z ⇌ K2Z + 2H2O(g) equilibrium is not unusual thermodynamically. It is the kinetics of K2(H2O)2Z dehydration and K2Z rehydration that are unusual. The K2Z lattice rapidly and reversibly expands by ca. 11% in a matter of minutes at room temperature to accommodate the two H2O molecules, a change equal to 18.0 Å3 per H2O at the 110 or 120 K temperatures of the SC-XRD structures and probably equal to ca. 20 Å3 per H2O at room temperature. For comparison, the average effective volume of a H2O molecule in a wide variety of alkali-metal and alkaline-earth salt hydrates is 24.5 Å3 (the 34 individual values ranged from 20.4 to 28.9 Å3; the anions were halides, hydroxides, and various oxoanions).78−80 We can put into perspective the rapidity with which the K2Z ↔ K2(H2O)2Z transformations occur by a comparison with two other potassium hydrates of 2− fluoroanions. The dihydrate K2(H2O)2(SnMe2F4), which has virtually the same K2(μ-H2O)22+ core as K2(H2O)2Z (see Table S25), undergoes dehydration at a measurable rate only above 75 °C.81 The monohydrate K2(H2O)(AlF5), in which the H2O molecule bridges two K+ ions with K−O distances considerably longer than those in K2(H2O)2Z (see Table S26), undergoes dehydration at a measurable rate only above 60 °C and was not completely dehydrated, at a heating rate of 3 °C min−1, until the temperature reached 140 °C.82 Furthermore, the anhydrous salt K2(AlF5) required several days to rehydrate back to K2(H2O)(AlF5) at 25 °C. For these reasons, our focus in the remainder of this paper will be on the unusual rates, not the extents, of dehydration and rehydration of structurally characterized M2(H2O)nZ compounds that are not microporous. We use the term latent porosity to describe the behavior of a nonporous crystalline material that undergoes unusually rapid and reversible volume changes to accommodate the absorption/desorption of reactive gases at 25 ± 5 °C.1 III.B. Dehydration of Li2(H2O)4Z and LiK(H2O)4Z. TGA traces for these two compounds are shown in Figure 7. In contrast to the rapid partial or complete dehydration at 25 °C of LiK(H2O)4Z, Na2(H2O)nZ (n = 3, 4), K2(H2O)nZ (n = 2, 4),1 Rb2(H2O)2Z, and Cs2(H2O)Z (see below), the lithium salt hydrate Li2(H2O)4Z did not lose any significant amount of coordinated H2O molecules in dry N2 at 25 °C over nearly 3 h. It began to lose H2O above 40 °C and, with a heating rate of 1.5 °C min−1, was completely dehydrated at ca. 140 °C. The anhydrous salt Li2Z was rehydrated when it was exposed to N2 containing 13 Torr of H2O(g), as shown in Figure S15. For comparison, it was reported that the compound Li2(H2O)7(B12H12) began to lose coordinated H2O molecules as soon as it was heated above 25 °C, as shown in Figure S16.69

Figure 6. DFT structures of Cs2Z (top) and Cs2(H2O)Z (middle) and the SC-XRD structure of Cs2(H2O)Z (bottom; ref 2). The large spheres are B12 centroids.

Rb2Z and the DFT and SC-XRD structures of K2Z. A reasonable alternative structure with the same P212121 symmetry as that experimentally observed for Cs2(H2O)Z was found to be considerably lower in energy than the C2/c structure. In fact, the DFT structures of Cs2Z and Cs2(H2O)Z and the SC-XRD structure of Cs2(H2O)Z all have nearly the same P212121 structure with respect to the positions of the anions and cations, as shown in Figure 6. They all have a distorted variant of the B82-Ni2In-like structures of K2Z and Rb2Z, in which the pseudo-close-packed layers of B12 centroids (⊙) are not rigorously coplanar (the centroids deviate from planarity by ±0.07 Å). The results listed in Tables 3 and S16−S18 show that the Cs−F, Cs···⊙, and ⊙···⊙ distances and the distances between the ⊙ planes are quite similar for Cs2Z and Cs2(H2O)Z. For example, the perpendicular distances between the least-squares planes of anion centroids are 5.28, 5.21, and 5.09 Å for DFT Cs2Z, DFT Cs2(H2O)Z, and SC-XRD Cs2(H2O)Z, respectively. Nevertheless, there is a small, but finite, 5.2 Å3 expansion of the DFT Cs2Z lattice to accommodate each H2O molecule upon hydration to DFT Cs2(H2O)Z. 12030

DOI: 10.1021/acs.inorgchem.7b02081 Inorg. Chem. 2017, 56, 12023−12041

Article

Inorganic Chemistry

the sample was heated above 40 °C. At a constant heating rate of 3 °C min−1, the sample lost all of the remaining H2O molecules and was transformed, after ca. 60 min, into anhydrous LiKZ at 200 °C. Given the structure of LiK(H2O)4Z, there is no doubt that the loss of the first H2O molecule involves breaking a Li−OH bond as well as a K−OH2 bond. Rehydration of anhydrous LiKZ was not studied because of the possibility of phase separation into, for example, Li2Z and K2Z. This possibility will be studied by PXRD in our future work. III.C. Rapid Room-Temperature Partial Dehydration/ Rehydration of Na2(H2O)4Z. We recently published the structure and dehydration behavior of Na2(H2O)4Z.7 The dehydration TGA is shown in Figure 8. The tetrahydrate phase rapidly lost

Figure 7. TGA plots showing the complete dehydration of Li2(H2O)4(B12F12) to Li2(B12F12) (top) and of LiK(H2O)4(B12F12) to LiK(B12F12) (bottom). In the latter experiment, the sample had lost a small amount of H2O at 25 °C after it was weighed but before the TGA data collection began. 1H, 11B{19F}, and 19F{11B} NMR spectra of the anhydrous compounds Li2(B12F12) and LiK(B12F12), taken after heating, showed that only a negligible amount of H2O was present (possibly due to adventitious H2O from the glovebox atmosphere) and that no observable degradation of the Z2− cluster had occurred.

At 68 °C, the composition was Li2(H2O)4(B12H12), which did not lose additional H2O molecules until it was heated above 105 °C. The anhydrous salt Li2(B12H12) was formed at ca. 160 °C69 and was stable until 250 °C, after which it decomposed with the evolution of H2.83 (Note that Li2Z is stable until heated above 400 °C.41) The compound Li(H2O)4(B(OH)4)·2H2O72 lost the four H2O molecules coordinated to the Li+ ion between 60 and 120 °C.84 The compound Li4(H2O)3(B8O13(OH)2), in which each of the three H2O molecules is coordinated in a terminal fashion to three of the four Li+ ions, loses H2O in stages beginning at 100 °C.85 The compound Li2Mg(H2O)4(ZrF6)2, with square-pyramidal Li(H2O)F4 coordination spheres, lost all four of the coordinated H2O molecules in one stage between 105 and 200 °C.63 As far as we are aware, there is no salt hydrate that rapidly evolves, in an inert gas at 25 °C, one or more H2O molecules that are coordinated only to Li+. In contrast to the behavior of Li2(H2O)4Z, the mixed-metal salt hydrate LiK(H2O)4Z lost 1.0−1.2 equiv of H2O in ca. 60 min at 25 °C. Mass loss slowed considerably after that until

Figure 8. TGA experiments with Na2(H2O)4Z (Z2− = B12F122−). The upper experiment shows the rapid dehydration of Na2(H2O)4Z to Na2(H2O)2Z in dry N2 in ca. 30 min at 25 °C, the resistance of Na2(H2O)2Z to further dehydration at 25 °C, and its complete dehydration at temperatures above 70 °C. The lower experiment, performed at a constant temperature of 23 °C, shows that Na2(H2O)2Z rapidly absorbs only a small amount of H2O under 14 Torr of H2O(g), absorbs 1.0 equiv of H2O in ca. 70 min under 18 Torr of H2O, and absorbs additional H2O very slowly after the composition has exceeded Na2(H2O)3Z.

2.0 equiv of H2O in a flowing dry N2 atmosphere in ca. 30 min at 25 °C, with no indication of the intermediacy of a trihydrate phase. Repeated attempts to crystallize and obtain the structure of Na2(H2O)2Z under a variety of conditions only resulted in the crystallization of Na2(H2O)3Z or Na2(H2O)4Z. The dihydrate 12031

DOI: 10.1021/acs.inorgchem.7b02081 Inorg. Chem. 2017, 56, 12023−12041

Article

Inorganic Chemistry

when the sample mass had returned to within 1% Na2(H2O)2Z. The first cycle was complete in ca. 90 min, and the subsequent cycles were complete in ca. 70 min. It can therefore be said that Na2(H2O)2Z exhibits latent porosity. Each 18 Torr of H2O(g) rehydration in Figure 9 began with a very rapid mass increase of ca. 0.05 equiv of H2O in ca. 1 min, which was followed by a much slower mass increase during the next 10 min, which was followed by the bulk of the hydration from Na2(H2O)2Z to “Na2(H2O)3Z” over the next 30−35 min. As in our previous latent porosity study,1 we define a pair of kinetic figures of merit that are useful for comparing the results of different hydration/dehydration cycles: the maximum rate of the bulk of the hydration of a particular sample, rhmax, and the maximum rate of the bulk of the dehydration of the same sample at the same temperature, rdmax. They are the absolute values of the maximum slopes of the hydration and dehydration TGA segments and therefore are easy to determine. They can be expressed in mg min−1 or %(total mass change) min−1. One of their uses is to determine the reproducibility of the rates of hydration and dehydration for consecutive cycles with a given sample. For example, for the three Na2(H2O)2Z ↔ “Na2(H2O)3Z” cycles shown in Figure 9, rhmax was 3.4, 3.8, and 4.0%(change) min−1 and rdmax for all three cycles was 6.6%(change) min−1. However, rhmax, rdmax, and the ratio rhmax/rdmax, which is dimensionless, are not intended to be used in any mathematical equation related to a mechanistic model or an equilibrium constant. As discussed in more detail below, we are not in a position to propose and test possible mechanisms of hydration and dehydration in this study. Nevertheless, anticipating a more detailed kinetic/mechanistic study in future work, we point out that the shape of the TGA hydration curves following the (presumed) rapid surface hydration resembles nucleation and growth models that have been proposed for many solid-state reactions, phase transformations, and crystallizations,86,87 including salt hydrate dehydrations and rehydrations.88,89 As we shall see, the shapes of the Na2(H2O)2Z hydration TGA curves also resemble the shapes of the hydration curves for isothermal TGA experiments with microcrystalline samples of K2Z, Rb2Z, and Cs2Z, which will be discussed next. III.D. Latent Porosity in K2Z, Rb2Z, and Cs2Z. The 10−15 mg samples of these compounds that were used in TGA hydration/ dehydration cycle experiments were finely ground for 10+ min in an agate mortar and pestle. Scanning electron micrographs of samples of K2Z and Cs2Z are shown in Figure S18. A majority of the individual particles are blocks or plates for which the largest dimension is ca. 1 μm or less. There are also some larger particles with dimensions up to ca. 5 μm, as well as conglomerates of many small particles that may or may not be welded together. As in our previous study of K2Z hydration/dehydration at 25 °C,1 we will not attempt to fit the TGA data to one or more specific solid-state reaction kinetic functions at this time because many important variables such as the particle size uniformity, particle-surface-area-to-volume ratio, crystallinity, surface roughness, etc., were not controlled in the experiments reported here. There are a large number of “mechanisms” (i.e., kinetic functions) by which hydrates as simple as monohydrates have been proposed to undergo dehydration, and one such kinetic function, or even a small number of possible functions, can rarely be assigned unambiguously because different functions usually fit the (almost always nonisothermal) mass change versus time data equally well.89−93

phase, Na2(H2O)2Z, did not lose any measurable amount of H2O at 25 °C, even after several hours. Heating the sample to 150 °C did result in the formation of anhydrous Na2Z. Direct comparisons of the similar structures and similar TGA dehydration behavior of Na(H2O)4Z and LiK(H2O)4Z are shown in Figure S17. The rehydration of Na2(H2O)2Z at 23 °C in N2-containing H2O(g) is also shown in Figure 8. Bulk rehydration did not occur when P(H2O) was 14 Torr but did occur rapidly at 18 Torr. With P(H2O) = 14 Torr, a small mass increase (ca. 0.05 equiv of H2O) commensurate with, presumably, surface rehydration, occurred in less than 1 min, but thereafter the mass did not increase even after an extended period of time (not shown). When P(H2O) was changed from 14 to 18 Torr, the mass increased somewhat beyond the composition “Na2(H2O)3Z” in only 70 min (from 60 to 130 min in the particular isothermal experiment shown in the bottom half of Figure 8) but then tapered off significantly. The transformation from “Na2(H2O)3Z” to Na2(H2O)4Z took an additional 24 h (not shown). The use of quotation marks is intended to indicate a particular Na2(H2O)nZ composition with n ≈ 3, not necessarily a specific trihydrate phase. Nevertheless, the shape of the Na2(H2O)2Z rehydration curve indicates that the transformation of Na2(H2O)2Z to Na2(H2O)4Z under 18 Torr of H2O(g) almost certainly involves the intermediacy of at least one Na2(H2O)nZ phase where 3 ≤ n < 4. Even if we assume that only one intermediate phase is present, it is not known if n is 3.0 or some fractional number between 3 and 4. In fact, the TGA plot suggests that n > 3. However, even if n = 3.0, the structure of the putative intermediate phase would not necessarily be the same as the P212121 polymorph of Na2(H2O)3Z determined in this work by SC-XRD and shown in Figures 3 and 4. In ongoing work, we plan to address these questions by collecting time-resolved PXRD data during Na2(H2O)2Z rehydration. Regardless of the structure of the Na2(H2O)nZ intermediate phase, we determined that it is dehydrated in dry N2 even faster than it is formed from Na2(H2O)2Z under 18 Torr of H2O(g). Three complete cycles carried out at 23 °C are shown in Figure 9. The carrier gas was changed from N2/18 Torr of H2O(g) to dry N2 when the sample composition just exceeded “Na2(H2O)3Z” and from dry N2 to N2/18 Torr of H2O(g)

Figure 9. Three complete Na2(H2O)2 ↔ “Na2(H2O)3Z” hydration/ dehydration cycles at 23 °C. 12032

DOI: 10.1021/acs.inorgchem.7b02081 Inorg. Chem. 2017, 56, 12023−12041

Article

Inorganic Chemistry

Figure 11. TGA hydration/dehydration cycles for a single sample of Rb2Z at 23.0, 25.0, and 27.0 °C. The carrier gas for the hydration segments was N2 containing 18 Torr of H2O(g). The carrier gas for the dehydration segments was dry N2. The carrier gas was changed ca. 0.5 min before the increase or decrease in mass in each segment was observed. The horizontal dotted lines correspond to the masses of Rb2Z and Rb2(H2O)2Z. The superimposed step-function plot shows the temperature changes from 23 to 25 to 27 to 25 to 23 °C before and after each of the six hydration/dehydration cycles.

for Rb2Z are shown in Figure 11. Several features are noteworthy. First, the fully hydrated masses at room temperature (23−24 °C) correspond, to better than ±0.7%, to the stoichiometries of the structurally characterized hydrates K2(H2O)2Z,1 Rb2(H2O)2Z (this work), and Cs2(H2O)Z2 (i.e., typically to better than ±5 μg starting with a ca. 13 mg sample of anhydrous salt). For example, the ratio of masses for the first and last hydrations of Rb2Z to Rb2(H2O)2Z at 23 °C shown in Figure 11 are 0.9366 (13.181 mg/14.072 mg) and 0.9365 (13.181 mg/14.074 mg), respectively, whereas the ratio of the two molar masses is 0.9362 [(528.648 g mol−1)/(564.679 g mol−1)]. Second, each hydration and dehydration segment begins with a rapid (ca. 1−2 min) increase or decrease in mass, respectively, that corresponds to, at most, a few percent of the total mass change, which, as discussed above, is tentatively attributed to the formation of a hydrated or dehydrated surface layer or layers. This is followed by a sigmoid-shaped region that constitutes the bulk of the mass change. Third, the previously defined rhmax and rdmax values occur when the sample is ca. 50% hydrated or dehydrated, respectively. For example, for the 24.0 °C Cs2Z ↔ Cs2(H2O)Z TGA shown in Figure 10, rhmax was 0.0451 mg min−1 (10.8% min−1) and rdmax was 0.00791 mg min−1 (1.9% min−1). Fourth, for a given sample, the hydration and dehydration TGA segments for multiple cycles at the same temperature are virtually superimposable, even when cycles at other temperatures are run between them. This is shown for two consecutive 23.4 °C K2Z ↔ K2(H2O)2Z cycles in Figure S19. There are also three 23 °C and two 25 °C Rb2Z ↔ Rb(H2O)2Z cycles in Figure 10 that can be compared in this way. Fifth, the fully hydrated mass of each sample, but not the fully dehydrated mass, decreased slightly as the temperature increased. Sixth, rhmax and rdmax are strongly temperature-dependent. This is readily

Figure 10. TGA plots of hydration/dehydration experiments at two temperatures with K2Z (top) and Cs2Z (bottom). The carrier gas was changed ca. 1 min before the increase or decrease in mass (i.e., not at 0 min for the hydration segments). The same sample was used at both temperatures in each of the experiments.

Galwey94 and others86,95 have pointed out the erroneous way that the term reaction mechanism has been equated with a particular kinetic f unction in many thermal analyses of solid-state reactions, as opposed to the more chemically relevant, nonmathematical, and widely accepted definition of a reaction mechanism as the “complete sequence of all simple [chemical] steps through which reactants are converted into products”.94 These caveats notwithstanding, the TGA results presented and discussed here can be tentatively described in terms of the generic chemical processes of (i) rapid surface hydration or dehydration of a particle of Z2− salt, (ii) relatively slow formation of one or more hydrated or dehydrated phases on the surface of the particle, and (iii) rapid growth of the new phase(s) throughout the particle (i.e., bulk hydration or dehydration). Note that these descriptions are only intended to facilitate the discussion and to serve as tentative hypotheses to be tested in future work on the hydration/dehydration behavior of these and other Z2− salt hydrates. Complete isothermal hydration/dehydration cycles at two temperatures are shown in Figure 10 for the same sample of K2Z (23.4 and 29.0 °C) and for the same sample of Cs2Z (24.0 and 32.0 °C). Multiple cycles at 23.0, 25.0, and 27.0 °C 12033

DOI: 10.1021/acs.inorgchem.7b02081 Inorg. Chem. 2017, 56, 12023−12041

Article

Inorganic Chemistry

Table 4. Hydration and Dehydration TGA Figures of Merita

apparent in Figure 10. For both sets of TGA plots in this figure, rhmax > rdmax at the lower temperature (23.4 or 24.0 °C) and rhmax < rdmax at the higher temperature (29.0 or 32.0 °C). The individual values are listed in Tables S27−S29, and plots of the ratio rhmax/rdmax versus temperature are shown for all three salt hydrate pairs in Figure 12. Room-temperature figures of merit for

temperature, °C dehydrated formula dehydrated mass, mg hydrated formula hydrated mass, mg mass change, mg rhmax, mg min−1 rhmax, %(change) min−1 rdmax, mg min−1 rdmax, %(change) min−1 rhmax/rdmax ratio 98% hydration time, min 98% dehydration time, min dehydrated/hydrated mass ratio formula mass ratio

24.7 K2Z 9.655 K2(H2O)2Z 10.441 0.786 0.0589 7.4 0.0189 2.4 3.1 18 61 0.9247

25.0 Rb2Z 13.181 Rb2(H2O)2Z 14.073(1) 0.892(1) 0.0310b 3.5b 0.0576b 6.5b 0.54b,c 40 25 0.9366(1)

24.0 Cs2Z 14.338 Cs2(H2O)Z 14.756 0.418 0.0451 10.8 0.00791 1.9 5.7 16 76 0.9717

0.9237

0.9362

0.9719

a 2−

Z = B12F122−. A single sample was used to determine all of these parameters for each compound. The carrier gas for the hydration segments was N2 containing 18 Torr of H2O(g). The carrier gas for the dehydration segments was dry N2. Molar masses (g mol−1): K2Z, 435.909; K2(H2O)2Z, 471.940; Rb2Z, 528.648; Rb2(H2O)2Z, 564.679; Cs2Z, 623.524; Cs2(H2O)Z, 641.539. bThese values are for the first cycle. cSee Figure 12 for the consistency of this ratio over three hydration/dehydration cycles at 23.0 °C.

Figure 12. Temperature dependence of the ratio rhmax/rdmax for hydration/dehydration TGA experiments with microcrystalline samples of K2Z, Rb2Z, and Cs2Z. The carrier gas for the hydration segments was N2-containing 18 Torr of H2O(g). The carrier gas for the dehydration segments was dry N2. The individual rhmax and rdmax values are listed in Tables S27−S29, respectively. For each compound, a single sample was used at all of the different temperatures.

samples of all three compounds are listed in Table 4. Plots of rhmax versus temperature are shown in the Table of Contents graphic. Close inspection of Figure 11 reveals that the differences between the “fully hydrated” mass of Rb2(H2O)nZ (n ≤ 2) and the constant, 13.181 mg, fully dehydrated mass of Rb2Z were 0.893(1) mg at 23 °C [i.e., 99.3% of the theoretical value of 0.898 mg assuming the exact stoichiometry Rb2(H2O)2Z], 0.884(1) mg at 25 °C, and 0.874 mg at 27 °C. This means that Rb2Z absorbed only 99% as much H2O at 25 °C than at 23 °C and only 98% as much at 27 °C than at 23 °C. This behavior was also observed for hydration/dehydration cycles of a 9.655 mg sample of K2Z, as shown in Figure S20. The mass increase at 23.4 °C was 0.792 mg, which is 99.4% of the theoretical value of 0.797 mg assuming the exact stoichiometry K2(H2O)2Z. Between 23.4 and 32.0 °C, the “fully hydrated” mass decreased monotonically to 0.772 mg, 96.9% of the theoretical value. These observations are consistent with, but by no means prove, the presence of H2O vacancies in the maximally hydrated compounds, potentially even at ca. 23 °C, which presumably increase in number with increasing temperature. This behavior will be studied in more detail in an ongoing investigation. The possible presence of H2O vacancies may be one of the keys to understanding how it is possible for D2O(g) to replace the H2O molecules in the “fully hydrated” compound nearly as quickly as H2O(g) is absorbed by the anhydrous salt in the first place. This was reported for K2(H2O)2Z in ref 1 (see Figure S2). We have now determined that Rb2(H2O)2Z also undergoes complete H2O/D2O exchange rapidly when exposed to 16 Torr of D2O(g) at room temperature, as shown in Figure 13. The solid-state hydration reaction Rb2Z(s) + 2H2O(g) → Rb(H2O)2Z(s) and the solid-state exchange

Figure 13. TGA plot showing the hydration of Rb2Z to Rb2(H2O)2Z in 60 min in the presence of 18 Torr of H2O(g) (segment a) and the subsequent solid-state (i.e., gas−solid) H2O/D2O exchange reactions Rb2(H2O)2Z + 2D2O(g) → Rb(D2O)2Z + 2H2O(g) in 64 min in the presence of 16 Torr of D2O(g) (segment b) and Rb2(D2O)2Z + 2H2O(g) → Rb(H2O)2Z + 2D2O(g) in ca. 80−90 min in the presence of 18 Torr of H2O(g) (segment c). The inset is an expansion of the H2O/D2O exchange reactions. The temperature of the sample was held at 23.0 °C throughout the entire experiment.

reaction Rb2(H2O)2Z(s) + 2D2O(g) → Rb(D2O)2Z(s) + 2H2O(g) both occurred in ca. 1 h at room temperature. However, unlike the hydration reaction, the H2O/D2O exchange reaction does not involve a solid-state phase change. This means that the Rb-coordinated H2O molecules continuously diffuse through the fully hydrated Rb2(H2O)2Z lattice at room temperature (i) at least as fast as D2O(g) molecules replace the H2O molecules and (ii) at least as fast as the bulk of the growth of the Rb2(H2O)2Z(s) phase from the Rb2Z(s) phase occurred during the hydration reaction. [Note that a control experiment 12034

DOI: 10.1021/acs.inorgchem.7b02081 Inorg. Chem. 2017, 56, 12023−12041

Article

Inorganic Chemistry

not bonded to the Rb+ cations).105 This compound did not lose any H2O until heated above 133 °C, as shown in Figure S23.105 The compound Rb8(H2O)14(Ta6O19) did not begin to lose H2O until ca. 70 °C, as shown in Figure S24. The composition stabilized at Rb8(H2O)4(Ta6O19) at 110 °C and only began to lose H2O again at ca. 175 °C, becoming anhydrous Rb8(Ta6O19) at ca. 260 °C.106 Finally, the compound Rb6(H2O)4(Mo7O24), with both doubly and triply bridging H2O molecules, did not begin to lose H2O until heated to 100 °C and, at a heating rate of 10 °C min−1, did not become completely dehydrated until ca. 180 °C, as shown in Figure S25.107 There are fewer structurally characterized cesium salt hydrates for which TGA results are available. Figure S24 shows the TGA plot for Cs8(H2O)14(Ta6O19), which did not begin to lose H2O until it was heated above 60 °C.106 The same was true for Cs6(H2O)19(U24O75)108 and Cs2(H2O)2(HPO4),109 and the latter compound did not form anhydrous Cs2(HPO4) until it was heated to 180 °C.109 The compound Cs2(H2O)2(Si2O5) began to lose H2O at 80 °C and formed anhydrous Cs2(Si2O5) at ca. 200 °C.110 The compound Cs2(H2O)3(B4O5(OH)4) began to lose H2O only when it was heated above 100 °C.111 The only literature cesium hydrate that comes closest to exhibiting what may be latent porosity is Cs2(H2O)(Fe(CN)5(NO)).112 This compound apparently did not lose H2O when held at 25 °C but began to lose H2O soon after it was heated and had become anhydrous Cs2(Fe(CN)5(NO)) by the time the temperature reached 43 °C, as shown in Figure S26 (the heating rate was not reported).112 The X-ray structure of the anhydrous compound, but not of the hydrate, has been reported.113 III.F. A Closer Look at Latent Porosity. We have shown that K2Z, Rb2Z, and Cs2Z exhibit what we describe as latent porosity. The compounds K2(B12H12), Rb2(B12H12), and Cs2(B12H12) do not exhibit latent porosity with respect to absorbing H2O(g): these anhydrous compounds are crystallized from aqueous solution.70,114 Two modifications of Rb2(H2O)2(B12(OH)12)) have been reported, both with RbO10 coordination spheres.115,116 One contains [−Rb(μH2O)Rb(μ-H2O)−]∞ chains;115 the other has a terminal H2O ligand on each Rb+ ion.116 However, no dehydration results have yet been reported for either modification. Interestingly, the salt Cs2(B12Br12) appears to be microporous as far as hydration is concerned. The three compounds Cs2(B12Br12), Cs2(H2O)(B12Br12), and Cs2(H2O)2(B12Br12) all crystallize in the R3̅ space group and have nearly identical unit cell parameters and hence nearly identical formula unit volumes (they differ by less than 1%; see Table S31 for details).70,117 Therefore, unlike Cs2(B12F12), the Cs2(B12Br12) lattice does not expand and contract upon hydration and dehydration, respectively. However, no observations about the rates of hydration of Cs2(B12Br12) in the presence of H2O(g) or the dehydration of either hydrate in the absence of H2O(g) have been reported. This should be investigated in the future. At the risk of confounding kinetics with thermodynamics, let us consider the following questions. To what can we attribute the latent porosity of K2Z, Rb2Z, and Cs2Z, behavior that may be all but unprecedented as far as metal salts that are not microporous are concerned? What allows H2O molecules to break their M−OH2 bonds and O−H···F hydrogen bonds so quickly as well as dif f use through the respective hydrate lattice so quickly? It is not because M−OH2 bonds are especially weak. As mentioned above, the enthalpy of dissociation of H2O from K2(H2O)2Z is 55.5 kJ (mol of H2O)−1,1 and this is not an unusually low value. In fact, the superweak nature of the Z2−

with K2(H218O)2Z and H2O, reported in ref 1, ruled out the possibility that the observed rapid H2O/D2O exchange involved proton transfer instead of intact isotopically labeled water molecules rapidly diffusing through the lattice and replacing H2O molecules.] III.E. Comparisons with Literature K+, Rb+, and Cs+ Salt Hydrates. As discussed in ref 1, the latent porosity behavior of K2Z is highly unusual, if not unprecedented, with respect to the behavior of other potassium salt hydrates, which require higher temperatures and/or significantly longer times to lose their coordinated H2O molecules. Comparisons with K2(H2O)2(SnMe2F2) and K2(H2O)AlF5 were discussed in section III.A. Comparisons with K(H2O)MnPO4, K(H2O)Zn2.5V2O7(OH)2, and a microporous synthetic potassium gallosilicate natrolite were discussed in ref 1. Another example from the literature is K4(H2O)4(P2S6),96 with KO2S6 and KO4S4 coordination spheres, which began to evolve H2O only at 50 °C and, with a heating rate of 5 °C min−1, did not form anhydrous K4(P2S6)97 until the temperature was 68 °C, as shown in Figure S21. Although neither TGA nor DSC/DTA results have been reported for the sructurally characterized compound K2(H2O)0.5(Pt(NO3)4), its crystals are “stable for a couple of weeks when kept in a desiccator protected from light”.98 Finally, although its structure is not known, the compound K2(H2O)2(Fe(CN)5(NO)) did not lose H2O until it was heated above 25 °C, displaying two DTA peaks associated with H2O loss at 64 and 96 °C, as shown in Figure S22.99 The rapid room-temperature latent porosity behavior we now report for Rb2Z and Cs2Z also appears to have no recognized parallel in the literature. Table S30 is a list of bond distances and bond valences for Rb2(H2O)2Z and seven other structurally characterized rubidium salt hydrates for which TGA results are available. The compound Rb4(H2O)6(P2S6) apparently did not lose H2O readily because it “was recrystallized from water at 60 °C placing the solution in a vacuum desiccator”.96 The compound Rb2(H2O)1.5(B4O5(OH)4)), with Rb(H2O)O9 and Rb(H2O)2O6 coordination spheres, two types of Rb(μ-H2O)Rb bridges, and rather long Rb−OH2 distances of 3.123(4)− 3.392(4) Å, did not lose any significant amount of H2O until heated above 100 °C.100 The H2O molecules in Rb2(H2O)2(HPO4) bridge four Rb+ ions, which have either Rb(H2O)4O4 or Rb(H2O)4O8 coordination spheres, with Rb−OH2 distances of 2.943(2)−3.178(2) Å.101 This compound did not lose H2O under vacuum at room temperature, even after several days.101 Anhydrous Rb2(Fe(CN)5NO) was formed, and a DTA peak was observed at 67 °C, when the monohydrate Rb2(H2O)(Fe(CN)5NO) was heated at 5 °C min−1 (the H2O molecule bridges the two types of Rb+ ions, which have RbON6 and RbON7 coordination spheres).102 The hydrated rubidium/molybdenum bronze Rb(H2O)(MoO3)4.55 did not lose H2O under vacuum for 48 h at room temperature and lost one H2O molecule per Rb+ ion at 70 °C.103 The compound Rb(H2O)(HMTA)I (HMTA = hexamethylenetetramine) has an Rb2(H2O)2 diamond-shaped moiety similar to the one in Rb2(H2O)2Z, with Rb−O distances of 2.898(4) and 2.963(4) Å, an O−Rb−O angle of 92.7(1)°, and an Rb−O−Rb angle of 87.3(1)°.104 This compound lost the coordinated H2O molecule only when heated above 58 °C.104 The compound Rb(H2O)2(LH4) has two Rb(H2O)O8 coordination spheres with terminal Rb−OH2 bond distances of 2.837(2) and 3.056(2) Å (LH4 is the monoanion of 4-amino-1hydroxybutylidine-1,1-bisphosphonic acid; one of the H2O molecules in the formula unit is a lattice H2O molecule and is 12035

DOI: 10.1021/acs.inorgchem.7b02081 Inorg. Chem. 2017, 56, 12023−12041

Article

Inorganic Chemistry

8% shorter than the distance within the rhombs. Note that there are [−Cs(μ-H2O)Cs(μ-H2O)−]∞ infinite chains running through the lattice in Cs2(H2O)Z.2 Finally, none of the H2O molecules in LiK(H 2 O) 4 Z, Na 2 (H 2 O) 4 Z, K 2 (H 2 O) 2 Z, Rb2(H2O)2Z, and Cs2(H2O)Z is hydrogen-bonded to another H2O molecule in the lattice, The H2O ligands are only weakly hydrogen-bonded to B−F bonds of the anions. It remains to be seen what effect the O−H···O hydrogen-bonded lattice H2O molecule in Na2(H2O)3Z has on the rate of dehydration or further hydration at room temperature. Table 5 lists structural results and information about the rates of dehydration for selected K+, Rb+, and Cs+ salt hydrates. Neither the M+/H2O ratio, the M+/anion ratio, the lattice volume per M+ ion, the lattice volume change per H2O molecule that is desorbed from the lattice, nor the percent of the total metal-ion bond valence due to the M−OH2 bonds can explain the rapid and reversible room-temperature dehydration of K2(H2O)2Z, Rb2(H2O)2Z, and Cs2(H2O)Z relative to the other salt hydrates. The only significant difference is the lack of O−H···O hydrogen bonding in the three Z2− compounds and its presence in the structures of most of the other compounds [the exceptions are K2(H2O)0.5(Pt(NO3)4), in which the H2O molecule bridges four K+ ions, and both K2(H2O) (Fe(CN)5(NO)) and Rb2(H2O)(Fe(CN)5(NO)), in which each H2O ligand bridges two K+ or Rb+ ions and participates in two O−H···N hydrogen bonds]. The latent porosity we have observed may be related to the SC-XRD and/or DFT structures of K2Z, Rb2Z, Cs2Z, K2(H2O)2Z, Rb2(H2O)2Z, and Cs2(H2O)Z, three of which are

anion might be expected to make metal ions only slowly give up their coordinated H2O ligands. The metal-ion coordination spheres in K2(H2O)2Z and Rb2(H2O)2Z are KO2F6 and RbO2F8, and 25% of the eight K−X bonds and 20% of the 10 Rb−X bonds (i.e., the two K−OH2 and Rb−OH2 bonds) account for 35% and 36% of the total bond valence of the metal ions, respectively (X = O, F). By this criterion, the average Rb−OH2 bond is about twice as strong as the average Rb−F bond in Rb2(H2O)2Z. It may be significant that Li2(H2O)4Z does not lose any amount of H2O at temperatures lower than 40−50 °C but that both LiK(H2O)4Z and Na2(H2O)4Z lose one or two H2O molecules in 120 or 30 min, respectively, at room temperature. The four H2O molecules in Li2(H2O)4Z are trapped in the center of a cage of Li+ and Z2− ions and are strongly hydrogenbonded into an (H2O)4 cluster. In addition, unlike the H2O molecules in LiK(H2O)4Z and Na2(H2O)4Z, the H2O molecules in Li2(H2O)4Z do not bridge two metal ions, precluding a relatively low activation energy path for diffusion of H2O through the lattice. There are no infinite chains of metal ions and bridging H2O molecules in K2(H2O)2Z and Rb2(H2O)2Z (the K···K and Rb···Rb distances in the K2(μ-H2O)2 and Rb2(μ-H2O)2 rhombs are 4.253 and 4.797 Å, respectively). Nevertheless, the H2O/D2O exchange reactions show that H2O diffusion through the hydrated salt lattice, which must involve the transfer of H2O molecules between rhombs, is rapid. This may be because the next-shortest K···K distance, which is between two rhombs, is 5.075 Å, only 20% longer than 4.253 Å, and the shortest Rb···Rb distance between rhombs is 4.400 Å,

Table 5. Comparison of Structural and Dehydration Results for Selected K+, Rb+, and Cs+ Salt Hydrates compd a

K2(H2O)4(B12F12) K2(H2O)2(B12F12)a K2(H2O)0.5(Pt(NO3)4)c K2(H2O)(Fe(CN)5(NO))d K4(H2O)4(P2S6)f Rb2(H2O)2(B12F12)h Rb4(H2O)6(P2S6)i Rb6(H2O)4(Mo7O24)j Rb8(H2O)14(Ta6O19)k Rb8(H2O)4(Ta6O19)k Rb2(H2O)(LH4)·H2Ol Rb2(H2O)(Fe(CN)5(NO))n Cs2(H2O)(B12F12)p Cs4(H2O)6(P2S6)i Cs8(H2O)14(Ta6O19)k Cs2(H2O)(Fe(CN)5(NO))

M+/H2O

M+/anion

V(M+), Å3

V(H2O), Å3

% bv from M−OH2

rapid dehydration at 23−24 °C

1:2 1:1 4:1 2:1 1:1 1:1 2:3 3:2 4:7 2:1 2:1 2:1 2:1 2:3 4:7 2:1

2:1 2:1 2:1 2:1 4:1 2:1 4:1 6:1 8:1 8:1 2:1 2:1 2:1 4:1 8:1 2:1

212.1 184.0 145.0 135.7 101.5 190.8 121.0 118.5 116.2 78.5 306.7 135.7 188.2 131.4 116.2 166.4t

23.1b 18.0b

50, 56 35 11 23, 28 33, 51 36 24, 36 13−31 45−62 17−39 8.3, 22 17 25q 22, 35 45−57

yes yes noc noe nof yesh noi noj nok nok nom non yesq noi nos nou

18.5g 16.4h 19.9g

16.6o 5.2q 21.5r

a

Reference 1; the M−OH2 bonds have the second and third highest bond valence (bv). bThe structure of anhydrous K2(B12F12) was published in ref 2. cReference 118; crystals were “stable for a couple of weeks when kept in a desiccator protected from light”. dReference 119; the structural results in this table are for the monohydrate. eReference 99; TGA/DTA plots indicate that dehydration of the dihydrate takes place in two stages, with DTA peaks at 64 and 96 °C. fReference 96; see Figure S21; dehydration occurred slowly between 50 and 68 °C upon heating at 5 °C min−1. g The structures of anhydrous K2(P2S6) and Rb2(P2S6) were published in ref 106. hThis work; the highest bv is for an M−OH2 bond. iReference 96; the compound “was recrystallized from water at 60 °C placing the solution in a vacuum desiccator”. jReference 107; no mass loss until heating above 100 °C. kReference 106; the loss of 10 H2O molecules from Rb8(H2O)14(Ta6O19) occurred between 68 and 100 °C; the loss of the remaining four H2O molecules occurred only above 175 °C (see Figure S24). lReference 105. mJunk, P. C. Personal communication, 2016 (see Figure S23). n Reference 102; dehydration produced “a DTA peak located at 67 °C”. oThe structure of anhydrous Rb2(Fe(CN)5(NO)) was published in ref 120. pReference 2; the highest bv is for an M−OH2 bond. qThis work; the sum of the Cs−OH2 bond valences was also 25% in the DFT-optimized structure. rThe structure of anhydrous Cs8(Ta6O19) was published in ref 106. sReference 106; the loss of 10 H2O molecules from Cs8(H2O)14(Ta6O19) occurred between 65 and 150 °C; the loss of the remaining four H2O molecules occurred between 150 and 200 °C (see Figure S24). tFrom the structure of anhydrous Cs2(Fe(CN)5(NO)), published in ref 113 (the structure of the hydrate is not known). u Reference 112; heating a sample resulted in a DTA peak at 43 °C (see Figure S26). 12036

DOI: 10.1021/acs.inorgchem.7b02081 Inorg. Chem. 2017, 56, 12023−12041

Article

Inorganic Chemistry

This so-called “negative kinetic temperature effect”, sometimes said to indicate “a negative activation energy”, is uncommon in condensed-phase chemical kinetics in general126−129 and is only rarely encountered in solid-state reactions.130−133 In solution, an apparent negative activation energy can be due to an equilibrium prior to the rate-determining step that shifts to the left with increasing temperature. In the case of the hydration of M2Z salts, it is possible that a surface hydration equilibrium controls the overall rate of hydration. This interesting phenomenon, and its possible causes, will be studied in an ongoing investigation. III.H. Thermal Stability of Anhydrous M2Z Salts (Z2− = B12F122−). Previously reported TGA experiments and beforeand-after 19F NMR spectra for Li2Z, K2Z, and Cs2Z demonstrated that they are stable with respect to cluster degradation until they are heated above 400, 500, and 600 °C, respectively,41 and that Na2Z is stable up to at least 327 °C.7 In this work, we have determined that the anhydrous salt Rb2Z is stable up to ca. 460 °C, as shown in Figure S27. We did not examine the thermal stability of LiKZ because of the ambiguity regarding possible phase separation into Li2Z and K2Z upon dehydration of LiK(H2O)4Z. For comparison, anhydrous LiPF6 decomposes at 133 °C,134 LiPF6 exposed to air [i.e., H2O(g)] at 91−105 °C135,136, LiBF4 at 162−243 °C135,136 (the wide range may also be a function of whether the salt is rigorously anhydrous), LiN(SO2F)2 at 200 °C,137 Li(1-MeCB11F11) at 300 °C,41 LiN(SO2CF3)2 at 340 °C,135 LiCF3SO3 at 425 °C,135 and K(B(3,5-C6H3(CF3)2)4) at 350 °C.138 We are not aware of a report about the thermal stability of a Rb+ salt of a weakly coordinating anion.

reported in this work for the first time. They are all based on a “slipped” variation of a hexagonal close-packed lattice of Z2− anions, and only relatively minor adjustments of anions within the close-packed layers and minor adjustments of the M···M distances are required for hydration to occur (see Table 3; the changes in the M···M distances upon hydration are 1.2 Å for K+, 0.8 Å for Rb+, and